Quote:
Originally Posted by rickroll
precept, i think you're running into the same issues i struggle with
for whatever reason, hard data in golf doesn't quite separate the wheat from the chaff, when i only look at shots gained stuff it can tell me crazy things so I need to weigh heavily with vegas odds and world rankings as a baseline metric to anchor people into place
unsure of your process though so hard to say exactly but when in doubt i average out with baseline rankings/vegas etc
Are you normalizing strokes gained data using field strength or just using the nominal/raw strokes gained data? Weeks like this could be case in point where you have a WGC and opposite field event with two distinctly different field compositions in terms of skill. This is a problem I want to tackle for my model because obviously data from weeks like this will skew a baseline skill determination.
Going 0SG across all categories at WGC is obviously not the same as 0SG across all categories at Barracuda. I just haven't figured out a way to approach this problem yet. Under normal circumstances there are 5-8 weeks a year where the tour plays two events and by not appropriately weighting SG you're going to likely overestimate skill for rookies/lower ranked guys since they are not competing against any of the top 40% of the tour that week.
World ranking does a decent job of at least attributing the idea of field strength, but it's well understood how there is bias in the ranking due its construction.